Spyke
feddit.nl

Technically possible with a small enough model to work from. It's going to be pretty shit, but "working".

Now, if we were to go further down in scale, I'm curious how/if a 700MB CD version would work.

Or how many 1.44MB floppies you would need for the actual program and smallest viable model.

44
Nazreply
sh.itjust.works

squints

That says , "PHILLIPS DVD+R"

So we're looking at a 4.7GB model, or just a hair under the tiniest, most incredibly optimized implementation of <INSERT_MODEL_NAME_HERE>

16
lemmy.dbzer0.com

llama 3 8b, phi 3 mini, Mistral, moondream 2, neural chat, starling, code llama, llama 2 uncensored, and llava would fit.

13
NoiseColorreply
lemmy.world

Might be a dvd. 70b ollama llm is like 1.5GB. So you could save many models on one dvd.

12
Ignotumreply
lemmy.world

70b model taking 1.5GB? So 0.02 bit per parameter?

Are you sure you're not thinking of a heavily quantised and compressed 7b model or something? Ollama llama3 70b is 40GB from what i can find, that's a lot of DVDs

8

Ah yes probably the Smaler version, your right. Still, a very good llm better than gpt 3

9

ELIZA was pretty impressive for the 1960s, as a chatbot for psychology.

7

Wow, I've actually never seen this disc. However, funnily enough, I have another disc with that program burned into it. Someone didn't read the notice, it seems.

3

Fun fact: you can download llama3, an llm model made by meta (which is surprisingly good for its size), and it's only 4.7gb. A dvd can store 4.7 gb of data, meaning you could in theory have an llm on a DVD.

Model link

CC BY-NC-SA 4.0

10
lemm.ee

Please keep us updated with all future inconsequential misinterpretations.

-Management

11

Dear Management,

I was just brushing my teeth without my glasses on, anf out of the corner of my eye it sorta looked like Versace.

Hope this helps, Facebones

3

You reached the end